You are on page 1of 25

Chapter 4

MIXING HEIGHT CALCULATION OVER NCT OF DELHI

4.1 Introduction
Transport and diffusion of pollutants in the lower atmosphere is dependent largely on the
structure of the planetary boundary layer (PBL), one important nature of which is the height of
the well-mixed layer. Massive quantities such as particles and gases are mixed nearly uniformly
throughout this layer by turbulence which results partially from strong surface heating during the
daytime hours. The mixing layer is capped by a temperature inversion thereby limiting the height
of the mixing. The variation of this height due to diurnal variations of solar radiation, synoptic
conditions and local terrain strongly affects pollutant concentrations.

4.2 AtmosplzericIPlanetary Boundary Layer


The height of the atmospheric boundary layer essentially governs vertical mixing of
atmospheric pollutants, which is expressed by the synonym "mixing height" with respect to
applications in environmental meteorology. It therefore plays an important role in air pollution
monitoring and assessment and serves as a basic input parameter to all classes of dispersion and
transport models.
In dispersion models, the mixing height is a key parameter needed to determine the
turbulent domain in which dispersion takes place. Air pollution climatology is concerned with
the aggregate of weather as it may affect the atmospheric concentrations of pollutants.
The mixing layer is the depth of the atmospheric layer which is characterized by strong
turbulent and convective mixing. Accurate representation of mixing depth plays an essential role
in the ability of models to predict pollutant concentrations (Vimont and Scire, 1994, Rao et aI.,
1994). There are several methods currently available for the estimation of mixing heights
(Holzworth, 1972; Garrett, 1981; Maughan et al., 1982; Myrick et al., 1994). They include those
using temperature profiles measured by sondes or by a fixed tower, and the remote sensing
techniques such as lidar and sodar (Coulter, 1979; Marsik et al., 1995). Each method has its own
advantages and limitations, and different methodologies give rise to differences in mixing
heights. These differences are the results of the physical limitations of each method and the

48
assumptions used as to which variable most accurately defines the depth of the mixed layer.
Among the most commonly methods of calculating mixing height from temperature profiles
are the Holzworth method (Holzworth, 1967), the capping method (Dayan et al., 1988), the
Heffter method (Heffter, 1980) and the Kink method (Goldman, 1980). The conventional
Holzworth mixing height is calculated as the intersection point of the dry adiabatic from the
surface temperature and the morning temperature sounding. It neither includes the effect of
horizontal temperature advection nor of moisture content and is sensitive to surface temperature.
Benkley and Schulman (1979) proposed an operational model based on Holzworth's met~od

including the effect of temperature advection. The daytime mixing height is normally identified
with the base of an elevated inversion or stable layer, capping the well-mixed convective
boundary layer. The capping mixing height is determined assuming that turbulent mixing caused
by convective and mechanical turbulence extends to the base of the elevated inversion. In Heffter
method, potential temperature profiles computed for each sounding are analyzed for the existence
of a critical inversion. The inversion is defined following the criteria that recognize the likelihood
of mixing to overshoot the base of the critical inversion. It does not inc~ude the effects of wind
shear within the critical inversion. The kink method estimates mixing height from vertical
temperature soundings with higher resolution than synoptic soundings. It assumes that vertical
mixing is limited at the base of the first significantly stable layer defined by imposed operational
conditions.
Remote ground-based observing systems, such as radar, sodar and lidar, have an advantage
over in situ measurements, their ability to obtain volume and continuous averages which are
more representative than point and instantaneous values. Radar and sodar see clearly the
convective boundary layer structure through the detection of small-scale variations in the
atmosphere's refractive index structure due to temperature and water-vapor fluctuations at the top
of the convective boundary layer. This result in a maximum backscatter signal intensity (Beyrich,
1995, Marsik et al., 1995). Radar systematically overestimates the depth of the mixed layer
during the early morning hours, and its measurements are affected by the presence of fog, clouds,
rain and persistent shear layers aloft (Marsik et aI., 1995). Sodar sends short pulses of sound, and
the attenuation of sound in the atmosphere causes difficulty in the detection of structures beyond
a range of about 1 km. This is an important limitation for operational application, specially in
summer, when many mixed layers grow to heights above that value in the afternoon (Stull, 1988;
Beyrich, 1995). When a neutral boundary layer exists, or no inversion layering is detected, no

49
information can be obtained. When multiple layers are observed, mixing height is taken to be the
height of the base of the lowest layer, but this could not be the strongest in terms of being a
barrier to vertical dispersion. The operation of a sodar is limited to relatively noise-free locations
and measurements are affected by wind and rain interface (Maughan et aI., 1982). Lidar transmits
laser light, which is scattered by air molecules, cloud droplets and aerosols in the boundary layer.
This method of mixing height determination measures the depth to which particles have become
mixed in the atmosphere (Stull, 1988; Maughan et al., 1982). The Lidar is less range limited in
the near field than other systems, and thus can detect the shallow, developing mixed layers during
the early morning hours. Lidar measurements are affected by clouds, lifting fog and elevated
plumes of aerosols and it does not operate during periods of precipitation (Marsik et aI., 1995).

Air pollution climatology is usually described in terms of statistics of wind, temperature,


stability, mixing height, ventilation coefficient, etc.
The climatological model CALMET applied in the present study gives the mixing height
along with visibility, stability, velocity and temperature fields for the study region. In the present
case, the mixing height determination uses an energy balance approach to estimate the surface
heat flux in driving the growth of the mixed layer (Scire et al., 1995) and the prameterizations
used in the energy budget are based primarily on Holtslag and van Ulden's (1993).
During the daytime, CALMET calculates the mixing height as the maximum of the
convective and mechanical mixing heights. The convective mixing height at time (t+l) is
estimated from time (t) in a stepwise manner as a function of sensible heat flux (Maul, 1980).
The mechanical mixing height is estimated from Venkatram's (1980a) relationship h= (BuJ ,
where u. is the friction velocity, f is the Coriolis parameter, and NB is the BruntWis~a
frequency in the stable air above.
During the nighttime, CALMET uses the minimum of the two niixing heights. The first
mixing height is estimated by Venkatram's (1980b) relationship h= B2 u. 3n., where
B2 = 2400 (constant). The second formula from Zilintinkevich (1972) is given as h= 0.4 (u.LIj) In. ,
where L is the Monin-Obukhov length.
In this chapter, the assimilative capacity (or carrying capacity, as often used
interchangeably) of the atmosphere is estimated. The assimilative capacity of the atmosphere is
the maximum amount of pollution load that can be discharged without violating the best designed
use of the air resources in the planning region. The phenomenon governing the assimilative

50
capacity of atmosphere includes dilution, dispersion, phase transformation, .deposition and
absorption.
The air pollution assimilation potential of an air shed, primarily for a plain terrain, can be
estimated as the ventilation coefficient which is an indicator of horizontal as well as vertical
mixing potential.
The ventilation coefficient over a region of atmosphere is the product of mixing height and
transport winds averaged over the entire mixing layer.
For the main purposes of the present study, the climatology of air pollution potential, the
non-precipitation data are considered to be adequately representative of all cases.

Estimation of assimilative capacity of the atmosphere involves:


Delineation of air shed based on topography of the area and identification of micro-
climatic zones depending upon sources, topography and wind field data.
Preparation of inventory ofpoint, area, and line sources, and quantification of pollution
loads.
Establishment of temporal and special variations of micro-meteorological parameters.
Prediction of temporal and special variations in air pollutant concentrations for existing
sources using multiple source-receptor model to establish source- receptor relationship.
Estimation ofavailable assimilative capacity in critical micro-climatic zones for various
pollutants vis-it-vis air quality standards for sensitive receptors.
Establishment of upper limits ofpollution load in critical pockets.

51
4.3 Model Description

4.3.1 Introduction
CALMET is a meteorological model which includes a diagnostic wind field generator
containing objective analysis and parameterized treatments of slope flows, kinematic terrain
effects, terrain blocking effects, and a divergence minimization procedure, and a
micrometeorological model for overland and overwater boundary layers.
The CALMET model contains Boundary Layer Model for application to overland grid cells,
over land surfaces, the energy balance method of Holtslag and van Ulden (1983) is used to
compute hourly gridded fields of the sensible heat flux, surface friction velocity, Monin-Obukhov
length, and convective velocity scale. Mixing heights are determined from the computed hourly
surface heat fluxes and observed temperature soundings using a modified Carson (1973) method
based on Maul (1980). Gridded fields of PGT stability class is also determined by the model. An
upwind-looking spatial averaging scheme is optionally applied to the mixing height and 3-
dimensional temperature fields in order to account for important advective effects.

4.3.2 Technical Description


The CALMET model uses a grid system consisting of NZ layers of NX by NY square
horizontal grid cells. Figure 4-1 illustrates one layer of grid cells for a 11 x 12 grid. The" grid
point" rerers to the center of the grid cell in both the horizontal and vertical dimensions. The
"cell face" refers to either the horizontal or vertical boundary between two adjacent cells. In
CALMET, the horizontal wind components (u and v) are defined at each grid point. The vertical
wind component (w) is defined at the vertical cell faces.

52
Grid Cell

G
R
I
D

C
E
L
L

I
N
D
E (XORIGKM, YORIGKM)
X X GRID CELL INDEX
Fig. 4.1. Schematic illustration of the CALMET horizontal grid system for a l1X12 grid showing
the grid origin (XORIGKM, YORIGKM) and grid point location (.)

The position of the meteorological grid in real space is detemlined by the reference
coordinates (XORIGKM, YORIGKM) of the southwest comer of grid cell (1,1). Thus, grid point
(1,1) is located at (XORIGKM + DGRIDKMl2., YORIGKM + DGRIDKMl2.), where
DGRIDKM is the length of one side of the grid square.

It is assumed that the orientation of the X and Y axes of the CALMET grid are west-east
and south-north, respectively. In this way, the grid system is compatible with the usual definition

53
of the u and v horizontal wind components as the easterly and northerly components of the wind,
respectively. One commonly-used grid system compatible with CALMET is the Universal
Transverse Mercator (UTM) Grid.
The CALMET model operates in a terrain-following vertical coordinate system.
(4.1)

where
Z is the terrain-following vertical coordinate (m),
z is the Cartesian vertical coordinate (m), and
hI is the terrain height (m).
The vertical velocity, W, in the terrain-following coordinate system is defined as:

(4.2)

where
w is the physical vertical wind component (m/s) in Cartesian coordinates, and
u,v are the horizontal wind components (m/s).

4.3.3 Divergence Minimization Procedure


Three-dimensional divergence in the wind field is minimized by a procedure (Goodin et
aI.1980), which iteratively adjusts the horizontal wind components (u,v) for a fixed vertical
velocity field so that at each grid point, the divergence is less than a user-specified maximum
value.

du dv dw (4.3)
-+-+-<e
dx dy dz

where
u,v are the horizontal wind components,
w is the vertical velocity in ten'ain following coordinates, and
E is the maximum allowable divergence.

In CALMET, the horizontal wind components are defined at the grid points. Vertical
velocities are defined at the vertical grid cell faces. Therefore, the divergence, D, at grid point

54
(i, j, k) is:

(4.4)

where Llx and Lly are the sizes of the grid cell in the x and y directions, respectively. For each
grid point, divergence is computed. The u and v wind components at the surrounding cells are
adjusted so that the divergence at the grid point is zero.The adjustments are:

'1/ \
. ~ new/i.IJ.k
1/.
,.IJ.k
1/ .
ad]
(4.5)

'1/ \
~ new/i_IJ.k
u.,- I J.. k - U d
aY
(4.6)

l
(Vnew IJ.I.k . v..
'J'
I k + V J.
aY
(4.7)

l
(Vnew iJ-I.k . v.. I k - V J.
'J- aY
(4.8)

where the adjustment velocities (uadj , vadj ) are:

-Dijkt::.X
U .-- (4.9)
aJ) 2

(4.1 0)

Each time the divergence is eliminated at a particular grid point, divergence is created at
surrounding points. However, by applying the procedure iteratively, the divergence is gradually
reduced below the threshold value, E, throughout the grid.

4.3.4 Micrometeorological Model

4.3.4.1 Surface Heat and Momentum Flux Parameters


Significant advances have been made in recent years In the understanding and
characterization of the structure of the planetary boundary layer (PBL) (Weil, 1985; Briggs,

55
1985). According to van Ulden and Holtslag (1985) and others, the use of the appropriate
boundary layer scaling parameters can improve the quality of dispersion predictions. The major
parameters needed to describe the boundary layer structure are the surface heat flux (Qh)' surface
momentum flux (p u/), and the boundary layer height (h). Several additional parameters,
including the friction velocity (u.), convective velocity scale (w.), and the Monin-Obukhov
length (L), are derived from these.
Hanna et al. (1986) have evaluated several models for the prediction of these boundary
layer parameters from routinely available meteorological observations. Two basic methods are
commonly used to estimate the surface heat and momentum fluxes. The first method known as
the profile method, requires at a minimum the measurement of the wind speed at one height and
the temperature difference between two heights in the surface layer, as well as knowledge of the
air temperature and roughness characteristics of the surface. Monin-Obukhov similarity theory
is then used to solve for the surface fluxes by iteration. The second approach, known as the
energy budget method, computes the surface heat flux by parameterizing the unknown terms of
the surface energy budget equation.

4.3.4.2 Overland Boundary Layer


An energy budget method, based primarily on Holtslag and van Ulden (1983), is used over
land surfaces in the CALMET micrometeorological model. The energy balance at the surface can
be written as:

(4.11 )

where
Q. is the net radiation (W/m2),
Qf is the anthropogenic heat flux (W/m2),
Qh is the sensible heat flux (W/m2),
Qe is the latent heat flux (W/m2), and,
Qg is the storage/soil heat flux term (W/m 2).

The ratio of the sensible heat flux to the latent heat flux is defined as the Bowen ratio.

(4.12)

56
The model will require gridded values of the Bowen ratio. The Bowen ratio is important
in determining the degree of convective turbulence because it reflects the partitioning of the
available energy into sensible and latent heat flux. Typical values of B range from'" 0.1 over
water bodies to ~ 10 for deserts.
The flux of heat into the soil or building materials, Qg' is usually parameterized during the
daytime in terms of the net radiation (Oke, 1978; Holtslag and van Ulden, 1983).

Qg cg Q. (4.13)

where the constant cg is a function of the properties of the surface. Oke (1982) suggests values
for cg of 0.05-0.25 for rural areas and 0.25-0.30 for urban areas. The larger values for urban areas
reflect the greater thennal conductivity and heat capacity of building materials. Holtslag and van
Ulden (1983) use a value of 0.1 for a grass covered surface.
The anthropogenic heat flux, Qf' is a function of the population density and per capita
energy usage. Oke (1978) summarizes annual and seasonally- averaged Qf values for several
urban areas. Although the Qf tenn has been retained for generality, it is usually small compared
to the other terms.
The net radiation, Q., is the residual of incoming (short-wave plus long-wave) radiation and
outgoing (long-wave) radiation. Q. can be expressed (Holtslag and van Ulden, 1983; Landsberg,
1981) as:

Q. Q".. (l - A) + Qlw-<l - Qlw-II (4.14)

57
where
Qsw is the incoming short-wave radiation (W/m2), consisting of a direct solar
radiation term (Qsw-s) plus a diffuse radiation term (Qsw-d),
A is the albedo of the surface,
Qlw-d is the incoming long-wave atmospheric radiation (W/m2), and,
QIW-u is the long-wave radiation (W/m2) emitted by the surface.

The method of Holtslag and van Ulden (1983) is used to estimate Q. The result of their
parameterization of each of the terms in eqn. (4.14) is:

(4.15)

(4.16)

where
T is the measured air temperature (deg. K),
(J is the Stefan-Boltzmann constant (5.67 x 10-8 W/m 2/deg. K4),
N is the fraction of the sky covered by clouds, and
<t> is the solar elevation angle (deg.).
Thelast term in eqn. (4.16) accounts for the reduction of incoming solar radiation due to the
presence of clouds. The values for the empirical constants c l, c2, c3, aI' ~, b l , and b2 suggested
by Holtslagand van Ulden (1983) are used (Table 4.1, APPENDIX - 4A). The solar elevation
angle is computed at the midpoint of each hour using equations described by Scire et al. (1984).
The Monin-Obukhov length is defined as:
3
L -PCp T u. (4.17)
kg Qh

58
where
T is the temperature CK),
cp is the specific heat of air at constant pressure (996 m 2/(s2. OK)},
p is the density of air (kg/m3), and
g is the acceleration due to gravity (m/s2).
Eqn. (4.16) is used to obtain an initial guess for u. assuming neutral conditions (L = 00).
This value of u. is used in eqn. (4.17) to estimate L. A new value for u. is then computed with
eqn. (4.17) and L. The procedure is repeated until convergence is obtained. Holtslag and van
Ulden (1983) suggest that three iterations are usually sufficient.
During stable conditions, WeiI and Brower (1983) compute u. with the following method
based on Venkatram (1980a):

u=CDN
- -UI[. C II2J (4.18)
. 2

(C ~ 0) (4.19)

u2 YZmg e, (4.20)
T
a _ _

where
CON is the neutral drag coefficient [k/ln(zm/zo)]'
y is a constant ('" 4.7), and,
Zm is the measurement height (m) of the wind speed, u.
The temperature scale, e., is computed as the minimum of two estimates:
(4.21)

The estimate of 8. is based on Holtslag and van Ulden (1982):

e. 1 0.09 (I - 0.5 N 2) (4.22)

and 8. 2 is:

59
(4.23)

The heatflux is related to u. and e. by:


(4.24)

and L is computed from eqn. (4.17).


The daytime mixing height is computed using a modified Carson (1973) method based on
Maul (1980). Knowing the hourly variation in the surface heat flux from eqn. (4.24) and the
vertical temperature profile from the twice-daily sounding data, the convective mixing height at
time t+dt can be estimated from its value at time t in a stepwise manner:

2 2 Qh (1 + E) dt 2 dB r hr]12 dBr.dr (4.25)


hr.dr hr + - + -

f "'1 P Cp "'1 "'1

2 E Qh dr]12
"'1 (4.26)
dB/.d,' - - ' - - - -
[ P Cp

where
ljT I is the potential temperature lapse rate in the layer above ~,
de is the temperature jump at the top of the mixed layer (K), and,
E is a constant ("" 0.15).
The potential temperature lapse rate is determined through a layer above the previous hour's
convective mixing height.
The neutral (mechanical) boundary layer height is estimated by Venkatram (1 980b ) as:

Bu
h - - ' - (4.27)
[! NBjl2

where
f is the Coriolis parameter ('" 10-4 S-I)
B is a constant ('" 2 112 ), and,
Ns is the Brunt-Vaisala frequency in the stable layer aloft.
The daytime mixing height could then be taken as the maximum of the convective and

60
mechanical values predicted by eqns. (4.25) and (4.27), however, such a procedure could cause
the resulting x-y field of mixing heights to have unreasonably large cell-to-cell variations, as each
grid cell's values of hI and h are computed independently.
In the stable boundary layer, mechanical turbulence production determines the vertical
extent of dispersion. Venkatram (1980a) provides the following empirical relationship to
estimate the stable mixing height.

(4.28)

where B2 is a constant ('" 2400).


The stable boundary layer height is estimated by Zilitinkevich (1972) as

(4.29)

CALMET defines the stable overland boundary layer height as the minimum of h) and h2
In the convective boundary layer, the appropriate velocity scale is W., which can be computed
directly from its definition using the results of eqns. (4.15) and (4.25).

(4.30)

where hI is the convective mixing height.

4.4 Input data required by CALMET

4.4.1 SUlface Meteorological Data: Hourly Observation of wind speed (mls), wind direction
(degree), temperature (K), cloud cover (tenths), ceiling height, (hundreds of fee!), surface
pressure (mb), relative humidity (percent), precipitation type code.

61
4.4.2 Upper Air Data: Twice-daily observed vertical profile of wind speed (mls), wind
direction (degree), temperature (K), pressure (mb), elevation (m).

4.4.3 Geophysical Data: Gridded fields of terrain elevations (USGS 1 km terrain elevation
resolution data set) , land use categories, surface roughness length, albedo, Bowen ratio, Soil
heat, flux constant, anthropogenic heat flux, vegetation leaf area index (fable 4.2, APPENDIX-
4B).

4.5 Experiment
The output data of ARPS model forms the input to the CALMET model by way of free
formatted files for surface data and in appropriate CALMET data format for upper air & geo-
physical data. The model was run for II-12th of January, April, August and October for the years
1991 to 1995. Since
.
there was no major disturbance during these months over the years, these
.

days were assumed to represent the Winter, Summer, Monsoon and Post-monsoon seasons
respectively. Missing values of temperature, cloud cover, ceiling height, surface pressure and
relative humidity are internally replaced by values at the nearest station with non-missing data.
Missing upper air wind speed, wind direction, or temperature are interpolated by CALMET. As
the program does not extrapolate upper air data, the top valid level must be at or above the model
domain and the lowest (surface) level of the sounding must be valid.
The model generated temperature field, wind field, mixing height, stability, Monin-Obukov
length etc. at the center of each grid point for the representative months over the periods '91 "to
'95. Since, the pattern of different fields are more or less the same for corresponding seasons over
the years, the simulated results for '95 only are discussed in this chapter. Out of the 132 grid point
values, two grids (8,8) and (7,2) were chosen (corresponding to warm pocket and cold pools
ofNCT of Delhi) and the diurnal variation of temperature, horizontal wind field, mixing height
and ventilation coefficients were plotted. Also stability classes and visibility range for different
seasons are plotted.

62
4. 6 Results and Discussion

Fig. 4.2 (a) - 4.2 (d) & Fig. 4.3 (a) - 4.3 (d) show the spatial variation of wind speed and
horizontal wind field respectively. The wind field is influenced by topography and temperature
anamoly at the surface level. The winds accelerated from south (cold pool) towards north (warm
pocket) ip. January (winter). Thus whichever grid is observed the wind direct towards warm
pocket. In April (summer) the winds which are relatively weak converged towards the warm
pocket in the north and diverged from the cold pool in the south. In the dense built up area winds
are further weakened (in the west). In the monsoon season, winds are strong and is generally
westerly/easterly depending upon the location of monsoon trough. In the present case the strong
westerly winds turned north-westerly/northerly, probably the coriolis force assisted in deviating
the wind direction. In the post-monsoon season winds are strong south-westerly deviating
towards east. Here too the winds originally blowing towards heat island in the north turned east
most likely due to coriolis effect.
On a rolling terrain with marginal variation, the effects on wind is found to be not
significant. So the plots of horizontal wind vector from CALl\1ET model simulation taking into
account the terrain features of the region is found to be the same as that obtained from the ARPS
simulated results discussed in the previous chapter.

Fig. 4.4 (a) - 4.4 (d) show the diurnal variation of simulated temperature at warm and cold grid
points for the representative months of '95. The simulation indicates an increase of temperature
from 00 GMT reaching its maximum value at 11 GMT both at urban and rural locations, then it
decreases slowly. In all the seasons this trend is more or less the same. In simulating the diurnal
variation, the radiative properties of the pollutants are not taken care of

Fig. 4.5 (a) - 4.5 (d) show the percentage frequency of different stability classes at warm and
cold pockets for the representative months of'95. The percentage frequency of stability classes
suggest, in the post-monsoon and winter months stable classes prevail. In summer more unstable
classes prevail. Due to strong winds and mixing, stability is more or less neutral most of the time
in monsoon.

63
5
4
4
3
3
2

1 2 3 4 5 6 7 8 9 10 11
1 2 3 4 5 6 7 8 9 10 11

(a) (b)

1 2 3 4 5 6 7 8 9 10 11
1 2 3 4 5 6 7 8 9 10 11

(d) (c)

Fig. 4.1 Numerical Simulation at 11 UTe of the Spatial variation of wind speed (m)
for the four representative months (in clockytise) of '95
.
....
... ..

.. /./ .. //./././.J ..
..... ' .
11~>:/

////Ijll 10 -;;':,'i:\:
/ / 1 / I / I If 9 ..... ~ ..... : .. : .... :
. , . .

III/III! / . '; i~/I , \ ,

"TII'rIII II 7,\';:/111:1"
... :

111
: ; ; ;

)'j"j-;-rfJ!! 8 ~\i~iirnr
5 ..... :..

4 ~ i.. ~ .. ~. :~ ...:, . ..:' .;


: ,... -: ..... I I I i 3 . !' , .... ,.:
.
. . . .. ..
. .
/"",i
3
2 .....1. . ( j.j.:.J...;.' ..... l ... 1 1 .
2 : : : : : :

1 ... { .. J .... ,. 1'-'1 '/'Ij+thlj"


00 1 2 3 4 5 8 7 8 9 10 11
00 1 2 3 4 5 6 7 8 9 10 11

(a) (b)

..
.. ..' .

/ . / / ../ /////
~: .JJ . J J[\. \..I) 1
....
...

1J .1.\.\\\ II
/////./
.././ /
. . . . .
9
/.
/./ //// //
.... .
8 JJJ \\\ lUI
L.J . .\,\.\\\1/.
.. /////////
... : : :. . 7 ....
777777777
..
..
.. ..
.. . 6 .. .1. :/.1:\.\.\111
"';/7/;';/7/77
IJl\VV\I/
5 ..
/'/////.//-
....... ........
" ", ..... ,' . '" ... . 4IlH\VVII I
////)..J-..J..._-
311 i\ :t \ \1 I I
/,~--- 2'1'1'\'\'\,\1"'1
1 ... t'1\\ '\"'\"\"""1""

00 1 2 3 4 5 6 7 8 9 10 11 00 1 2 3 4 5 8 7 8 9 10 11

(d) (c)

Fig. 4.3 Numerical Simulation at 11 UTe of the Spatial variation of horizontal wind field
for the four representotive months (in clockwise) of '95
285.5 299
285 298.5
284.5 298
g 284 g
...
CD 297.5
...:s 283.5
~
...!:s
~ 297
CD 283 CD
Q. Q.
E 282.5 E 296.5
.s .s 296
282
281.5 295.5
281 295
0 10 20 0 10 20
time (hour) time (hour)

299 302

301
298
g g 300

..
!
:s
~
CD
297 ..!
:s 299
~
Q. 8. 298
E E
.s S 297
296
296
295 295
0 10 20 0 10 20
time (hour) time (hour)

Fig. 4.4. Diurnal variation of temperature at warm & cold grid points
for the four representative months (in clockwise) of '95 . .
[ -;-- ------j _._----._-,
35 50
I!
>- 30 >- 45
u u
cQ) c
Q)
40
~
0"
25 ~
0" 35
....!!!
....0 20
!!!
........ 30
[;~ban 1
, 0 25
Q)
Cl 15 Q)
Cl 20
_~~!~LJ
I
2c 2c
10 15
...ellu
Q) ell
...
U
ell
I 10
C. 5 !
J
c. 5
I
I
0
2 3 4 5 6 2 3 4 5 6
PGT stability classes
I PGT stability classes I

--t-
I
--- ---------' :

I
50 80
45
>-
u u
>- 70
c 40 cQ) 60
ell
~
0"
35 ~
0"
....!!! 30 ....!!! 50
'0 25 ~u~;nll ....0 40
18-~~ba~-1
ell
Cl ~~ura~ J ! ell /!I~U!.a..I _1
....c"' 20 Cl
2c 30
15
... ...uell
ell Q)
U 20
ell 10
a. a.
5 10
0 0
2 3 4 5 6 2 3 4 5 6
PGT stability classes PGT stability classes i
_______ _..1I

Fig. 4.5. Percentage frequencies of stability at warm & cold grid points
for the four representative months (in clockwise) of '95.
Fig. 4.6 (a) - 4.6 (d) show the diurnal variation of Monin-Obukhov length at warm and cold
pockets for the representative months of 195. The diurnal variation of Monin-Obukhov length
further corroborates by registering more stable conditions from late night to early morning in
almost all the seasons with large duration during winter compared to other seasons.

Fig 4.7 (a) - 4.7 (d) show the diurnal variation of mixing height at warm and cold grid points for
the representative seasonal months of 195. As expected, the mixing height increases from
minimum in the morning to a maximum around 11 GMT and decreases thereafter. This response
is clearly due to diurnal variation in surface temperature. Winter months are characterised by low
values of mixing heights (both maximum and minimum). Summer and post monsoon months
show comparatively high value of maximum mixing heights. Monsoon months show high values
of morning mixing heights and comparatively low values of afternoon mixing heights. There is
a bodily shift of-these curves from urban to rural and vice-versa due to urban effects on the
radiative characteristics.

Table 4.3 show the minimum (3 GMT) and maximum (11 GMT) mixing heights and ventilation
coefficients for the representative seasonal months (mean of 191 to 195). Morning ventilation
coefficients are very low in all the seasons. According to Gross (1970) criteria for high pollution
potential, the morning mixing heights should be :::; 500m and transport wind :::; 4 mls and
afternoon ventilation coefficients should be :::; 6000m2/s and transport wind :::; 4m1s. Applying
these criteria to Delhi it is found that morning periods have high pollution potential in all seasons
and it decreases gradually with time by the afternoon.

4.8 Conclusion

1. The diurnal variation of temperatures at both the warm and cold pockets followed same
trend and attained peak values in the afternoons.
11. Mixing heights also exhibited similar trends as that of the temperature. Mixing heights
during winter are lowest in all the months. On the otherhand mixing height are highest
in summer and monsoon months. Urban mixing heights are relatively higher than rural
and the time of Occurrence is later at the latter location.

64
580 155
135
480
115
g g
,.
380 95
.s::.
oOJ
;;
CI CI
c 280 c 75
.!! .!!
0, q 55
::E 180 ::E
35
80

-20
0
-- 10
time (hour)
20
15

10
time (~our)
20

1925
480
1725
430
1525
380
1325
330
g g 1125
280 .s::.
.s::.
oOJ
oOJ
CI
CI
c 230 c 925
.!! .!!
q 180 q 725
::E ::E
130 525
80 325
30 125
-20 -75
0 10 20 0 10 20
time (hour) time (hour)

Fig. 4.6. Diurnal variation of Monin-Obukhov length at warm & cold grid points
for the four representative months (in clockwise) of '95.
-- - ~ ~ - _....._ -- -
600 800
700
500
g g 600
....
.J::.
400 .... 500
.J::.
C! C!
'iii 300 'CD 400
.J::. .J::.
C! C!
c: c:
'j( 200 'x 300
'E E 200
100
100
0
0 10 20 o. 10 20
time (hour) time (hour)

- --

1000 800
900 700
800
g
..... 700 E
....
600
.J::.
en 600 1: 500
en
'CD 500 Gi 400
.J::. .J::.
en
c: 400 en
c 300
'x 'j(
'E 300 'E 200
200
100 100
0 0
0 10 20 0 10 20
time (hour) time (hour)

--- - -- - -

Fig, 4}. Diurnal variation of mixing height at warm & cold grid points
. .. for the four representative months (in clockwise) of '95,
Table 4.3

Mean Mixing height (at 3 GMT & 11 GMT)


at {(S,S) grid point} NeT of Delhi.

Month/Season Mixing height


3 GMT 11 GMT

January (winter) 100 575


April (summer) 135 785
August ( monsoon) 185 750
October (post-monsoon) 225 850

Mean ventilation coefficients (at 3 GMT & 11 GMT)


at {(S,S) grid point} NeT of Delhi.

Month/Season Ventilation coefficient

3 GMT 11 GMT

January (winter) 500 6,000


April (summer) 2100 6,600
August (monsoon) 1700 6,300
October (post-monsoon) 1200 6,600
111. More stable periods occurred in post-monsoon and winter seasons but more unstable
periods are registered in summer. Monsoon season recorded more neutral stabilities.
IV. This is further supported by Monin-Obukhov length which showed more stable
conditions from late night to early morning - the duration varied with season.
v. Diurnal and seasonal ventilation coefficients suggest high pollution potential during
morning compared to afternoons and in winter compared to summer.

In the next chapter, a brief description about the puff apporoach in studing the diffusion of
pollutants , the Gaussian-puff model to study this effect and the assimilative capacity of the
atmosphere are discussed.

65